Computing non-equilibrium trajectories by a deep learning approach

DOI10.1016/J.JCP.2023.112349MaRDI QIDQ6095083zbMATH OpenFDO

Authors Eric Simonnet

Publication date 27 November 2023

Published in Journal of Computational Physics (Search for Journal in Brave)

Copyright license Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International

Full work available at URL https://arxiv.org/abs/2210.04042

zbMATH Keywords

Freidlin-Wentzell large deviation theorygeometric actiongMAMinstantonsneural networksquasi-potential

Mathematics Subject Classification ID

Stochastic analysis (60Hxx) Probabilistic methods, stochastic differential equations (65Cxx) Numerical methods for partial differential equations, initial value and time-dependent initial-boundary value problems (65Mxx)

Abstract: Predicting the occurence of rare and extreme events in complex systems is a well-known problem in non-equilibrium physics. These events can have huge impacts on human societies. New approaches have emerged in the last ten years, which better estimate tail distributions. They often use large deviation concepts without the need to perform heavy direct ensemble simulations. In particular, a well-known approach is to derive a minimum action principle and to find its minimizers. The analysis of rare reactive events in non-equilibrium systems without detailed balance is notoriously difficult either theoretically and computationally. They are described in the limit of small noise by the Freidlin-Wentzell action. We propose here a new method which minimizes the geometrical action instead using neural networks: it is called deep gMAM. It relies on a natural and simple machine-learning formulation of the classical gMAM approach. We give a detailed description of the method as well as many examples. These include bimodal switches in complex stochastic (partial) differential equations, quasi-potential estimates, and extreme events in Burgers turbulence.

Recommendations

• Predicting rare events using neural networks and short-trajectory data
• Using machine learning to predict extreme events in complex systems
• Numerical computation of rare events via large deviation theory
• The geometric minimum action method: A least action principle on the space of curves

Cites work

• A method for representing periodic functions and enforcing exactly periodic boundary conditions with deep neural networks
• A trace formula for activated escape in noisy maps
• Adaptive Multilevel Splitting for Rare Event Analysis
• Chaos, Scattering and Statistical Mechanics
• Collapse of transitional wall turbulence captured using a rare events algorithm
• Combinatorial analysis of the adaptive last particle method
• Computation of extreme heat waves in climate models using a large deviation algorithm
• Computation of minimum action paths of the stochastic nonlinear Schrödinger equation with dissipation
• Computing transition rates for the 1-D stochastic Ginzburg-Landau-Allen-Cahn equation for finite-amplitude noise with a rare event algorithm
• Coupling rare event algorithms with data-based learned committor functions using the analogue Markov chain
• Crises, sudden changes in chaotic attractors, and transient chaos
• Deep splitting method for parabolic PDEs
• Deterministic Nonperiodic Flow
• DGM: a deep learning algorithm for solving partial differential equations
• Finding the quasipotential for nongradient SDEs
• Fluctuations and Irreversible Process. II. Systems with Kinetic Energy
• Gel’fand–Yaglom type equations for calculating fluctuations around instantons in stochastic systems
• scientific article; zbMATH DE number 3826915 (Why is no real title available?)
• scientific article; zbMATH DE number 1239549 (Why is no real title available?)
• scientific article; zbMATH DE number 2106098 (Why is no real title available?)
• Hyperbolic periodic orbits in nongradient systems and small-noise-induced metastable transitions
• Instanton filtering for the stochastic Burgers equation
• Julia: a fresh approach to numerical computing
• Large fluctuations for a nonlinear heat equation with noise
• Minimum action method for the study of rare events
• Multi-scale deep neural network (MscaleDNN) for solving Poisson-Boltzmann equation in complex domains
• Physics-informed neural networks: a deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations
• Probability density functions in steady-state Burgers turbulence
• Simulation and estimation of extreme quantiles and extreme probabilities
• Solving eigenvalue PDEs of metastable diffusion processes using artificial neural networks
• Solving for high-dimensional committor functions using artificial neural networks
• Statistical behaviour of adaptive multilevel splitting algorithms in simple models
• Study of noise-induced transitions in the Lorenz system using the minimum action method
• Symmetry-increasing bifurcation of chaotic attractors
• The geometric minimum action method: A least action principle on the space of curves
• The instanton method and its numerical implementation in fluid mechanics
• Towards a theory of transition paths
• Unbiasedness of some generalized adaptive multilevel splitting algorithms
• Understanding and Mitigating Gradient Flow Pathologies in Physics-Informed Neural Networks
• Viscous instanton for Burgers' turbulence

Cited in

(3)

• Intelligent dissipative particle dynamics: bridging mesoscopic models from microscopic simulations via deep neural networks
• Predicting rare events using neural networks and short-trajectory data
• Large deviation minimizers for stochastic partial differential equations with degenerate noise

This page was built for publication: Computing non-equilibrium trajectories by a deep learning approach

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6095083)